Jake Brodsky, automation guru and SCADA gadfly, posted this on the SCADA list earlier today, and I asked his permission to post it.
Folks, I like radios. I've been fascinated with them since I was five years old. I tinker with them to this day as an amateur radio operator.HOWEVER, there is a place and time for everything. And in my opinion, unlicensed wireless gear is NOT suitable for plant control system infrastructure of any significance.
Here's why:
1) This is the most important reason: unlicensed operation is not protected in any way. You must not cause interference to anyone else and you must accept interference from them if it happens. If evidence of interference with licensed operations is demonstrated, you must cease all operation until you can find some way to avoid interfering with the licensed operation. That's the gist of what it says in the US Federal Communications Commission, regulations, part 15. Most countries have similar regulations on the books.
2) Many standards used by the wireless marketeers are designed for minimal power use (Zigbee in particular). However, because they use spread spectrum methods, the first stages of the receiver must have very little frequency selectivity. This means the receivers can pick up activity from well outside the ISM band they're assigned to, not just inside. There are lots of very powerful signal sources outside the ISM band as well as inside. This can overload a receiver's first stages. You need lots of power in the first stages of a receiver to avoid overloading and causing a phenomenon called Blocking. But where does that power come from when you're using batteries and the power budget is tight?
3) Many cite process gain as a way to avoid problems with other users of that spectrum. But what they don't tell you is that the process gain behavior is predicated on linear behavior in the receiver. As soon as the receiver is deluged with enough signal to make it non-linear (when blocking starts, for example), the process gain margins start to fall.
4) Furthermore, many advocates suggest that because of process gain and spreading, that this technology is nearly jam-proof. Ladies and Gentlemen, nothing could be further from the truth. Just ask folks who know a bit about GPS. It is a spread spectrum system and it gets jammed all the time. Most of jamming models I've seen in the context of SP-100 assume either white noise or coherent carrier attacks. But with a bit of tweaking, a pulsed signal can jam them very nicely. (I'm not going into details on how that works because that would take too much to describe, and this isn't the forum for such things.)
Finally, with most other wired networks, you can test the cable with an off-the-shelf test set with simple go/no-go indications. But how do you validate an RF path? How do you identify the problem and fix it? Keep in mind, when it fails, it will probably take your process with it.
Even if your instrument technicians realize the problem is an RF path issue, what do they do next? Do you have the skill and equipment to deal with a spread spectrum path problem? Can you figure out where the source of the problem is coming from? With all the spread spectrum emitters on the drawing board today, how will you know where to find the ones that bring you down?
Where does this leave the ISA's SP-100 standards? I don't know. I'll be happy to use these things for non-critical applications such as an extra DO probe in a waste water aeration basin. But for most process loops it ain't happening with me, and I don't recommend anyone try it.
Jake Brodsky
Leaders relevant to this article: